The effect of alkaline pretreatment on surfactant- modified clinoptilolite for diclofenac adsorption: isotherm, kinetic, and thermodynamic studies

The elimination of diclofenac traces from aqueous environments is important. In this research, the effect of alkaline (NaOH) pretreatment on clinoptilolite before its modification with a surfactant (HDTMA) for diclofenac adsorption under the speculation of the sole presence of diclofenac in the aqueous solution is investigated. The results are compared through isotherm, kinetic, and thermodynamic studies and supplemented by Fourier transform infrared spectroscopy (FTIR), scanning electronmicroscopy (SEM), Brunauer–Emmett–Teller (BET), and the zeta potential analyses. The contact time was investigated in a 0–180-min range. The pH effect was studied in a range of 5–10 because of diclofenac dissociation below pH1⁄4 5. The effect of the temperature on diclofenac adsorption was also considered by establishing the experiments at 25, 35, and 45 C. For HDTMA-modified clinoptilolite, Temkin, and for NaOH-HDTMA-modified clinoptilolite, Dubinin–Radushkevich, and Freundlich isotherm models and in both cases, the pseudo-second-order kinetic model fitted the experimental data best. All the enthalpy and the entropy changes were negative, suggesting exothermic adsorption with a decrease in the degree of freedom of diclofenac anions after the adsorption. Furthermore, diclofenac physisorption was confirmed through isotherm and kinetic studies.


INTRODUCTION
As the human population increased during recent years, the need for drinking water supplies grew. Some pharmaceuticals escape the wastewater treatments and pollute the rivers; they can remain in water resources and threaten the ecosystem (Akhtar et al. ). Diclofenac (DCF) is an acidic pharmaceutical which belongs to the non-steroidal pharmaceutical group with an analgesic effect. It is prescribed for arthritis, rheumatism, and after-surgery inflammations (Sotelo et al. ). DCF is one of the most common pollutants in aqueous environments. It can be degraded in surface waters by sunlight, but the formed products are more dangerous to other living organisms (Huguet et al. ).
Continuous uptake of DCF into the human body, even in low concentrations, causes body function failure (de Luna et al. ). The accumulation of DCF residues in human organs causes serious health problems such as kidney, liver, and tissue damages. The limit of detection of DCF is reported to be between 1 and 10 ng/L (Poorsharbaf Ghavi et al. ). Removal methods such as adsorption, oxidation, ozonation, and Fenton processes have been applied to eliminate pharmaceutical traces from the aqueous environments. Adsorption, however, is of more interest because of its simplicity and effectiveness (Hu & Cheng ). Zeolites are commonly used adsorbents because of their low cost and adsorptive characteristics. Zeolites have a net negative charge, so they have no or low affinity toward anions (Warchoł et al. ). Surfactants are molecules with long chains and have both hydrophobic and hydrophilic parts (Urum & Pekdemir ). In an aqueous solution, the surfactant forms aggregations at a specific concentration. This is the critical micelle concentration (CMC).
The surface ion exchange capacity of the zeolite increases as a result of zeolite modification by surfactants; thus, the surfactant-modified zeolite shows a better function in the adsorption of anions and organic compounds (Misaelides ). bilayers. If the initial concentration of HDTMA is less than its CMC, monolayers appear, and if it is more than the CMC, admicelles form on the adsorbent surface. These admicelles then rearrange and bilayers appear. In bilayers, the positive head of HDTMA tends to be toward the solution. This helps the modified adsorbent to adsorb anions (Ambrozova et al. ). Alkaline treatment with NaOH has proven to increase the cation exchange capacity of zeolite and if it is followed by a surfactant modification, the adsorbent effectively adsorbs organic compounds (Tran et al. ). Alkaline pretreatment can be used for other purposes such as the bioconversion of lignocellulosic biomass. This pretreatment is advantageous, as it can simply be carried out under ambient temperature, it is cheap, and the procedure is easy to exploit (Kim et al. ). The modification of clinoptilolite with NaOH increases its sodium content, and the presence of the sodium cations can improve the cation exchange capacity of clinoptilolite (Ates & Akgül ).
The literature review showed that HDTMA-modified clinoptilolite (HDTMA-Clino) was an effective adsorbent for DCF. On the other hand, alkaline pretreatment of a zeolite improved its characteristics as an adsorbent through the enhanced surface area. The combination of these two modification methods and the comparison between them for DCF adsorption were missing from previous studies, so this study focuses on DCF adsorption with HDTMA-Clino and NaOH-HDTMA-modified clinoptilolite (NaOH-HDTMA-Clino). The feasibility of HDTMA-Clino and NaOH-HDTMA-Clino for DCF adsorption in the aqueous solution is assessed under different experimental conditions, by varying the contact time, the pH of the solution, the adsorbent amount, the initial DCF concentration, and the temperature.
The experimental data are studied alongside with the results from different analyses of the adsorbents to reach reliable conclusions. To investigate the effect of alkaline pretreatment on the adsorbent characteristics, two modification conditions were adjusted. In the first set of conditions, 4 g of clinoptilolite was added to a 1.5 mol/L NaOH solution and mixed in an incubator shaker at 200 rpm and 25 C for 24 h. In the second set of conditions, 4 g of clinoptilolite was added to a 1.5 mol/L NaOH solution and mixed at 260 rpm and 55 C for 90 min. After the alkaline pretreatment, the adsorbent was washed with distilled water until the solution pH was near 7. Then, the NaOH-pretreated clinoptilolite was modified by HDTMA. DCF adsorption with NaOH-HDTMA-Clino prepared under these two sets of conditions was compared, and the proper set was selected.

MATERIALS AND METHODS
The effect of the contact time, the pH of the solution, the adsorbent amount, and the temperature on DCF adsorption was investigated. The removal percent and the adsorption capacity of HDTMA-Clino and NaOH-HDTMA-Clino for DCF adsorption were investigated at 20, 50, and 80 mg/L DCF solutions in a contact time range of 0-180 min. The effect of the solution pH was studied in a pH range of 5-10 for 50 mg/L DCF solutions and 0.5 g adsorbent in 100 mL volume. The effect of the adsorbent amount was studied with 0.5, 1, 1.5, and 2 g of HDTMA-Clino and NaOH-HDTMA-Clino for 50 mg/L DCF solutions. All the pH and the adsorbent amount investigation experiments were done at 25 C and 200 rpm, at the equilibrium time.
Also, the effect of DCF initial concentration and the temperature was investigated at three different temperatures: 25, 35, and 45 C. The obtained experimental data were adjusted to Langmuir, Freundlich, Temkin, and Dubinin-Radushkevich (D-R) isotherm models to find the isotherm that explained the adsorption best. During all the isotherm study experiments, the parameters were set in their optimum values, while the initial concentration and the temperature were the variables. To find the kinetic model that elucidated the DCF adsorption well, pseudo-first-order, pseudo-secondorder, and intraparticle diffusion models were fitted to the experimental data and their deviations from the experimental data were calculated.
The Fourier transform infrared spectroscopy (FTIR) analysis was done using a Nicolet 560 FTIR spectrometer.
A minimum amount of 50 mg of each powdered sample of clinoptilolite, HDTMA-Clino, and NaOH-HDTMA-Clino were prepared, and the FTIR spectra of these samples were obtained from 400 to 4,000 cm À1 to confirm the functional groups of the loaded surfactant. The scanning electron microscopy (SEM) images were taken by a Philips XL30 (WDX: WDX-3pc, Microspec) SEM device. The Brunauer-Emmett-Teller (BET) surface area, the total pore volume, and the average pore volume along with the related N 2 adsorption-desorption diagrams were obtained using a Belsorp mini II BET device for Microtrac Bel Corp. The zeta potential was determined by dispersing clinoptilolite, HDTMA-Clino, and NaOH-HDTMA-Clino in distilled water and using a ZetaCheck device for MicroTrac Company.
The removal percent was calculated with Equation (1) and the adsorption capacity was calculated with Equation (2): In Equations (1) and (2), C 0 represents the initial concentration of the adsorbate and C e represents the equilibrium concentration in terms of mg/L. Parameter m is the adsorbent amount in g, and V is the solution volume in L. %R shows the removal percent, and q e is the equilibrium adsorption capacity of the adsorbent in mg/g. Langmuir isotherm can be represented by the following equation: In Equation (3), q e is the equilibrium adsorption capacity and q m is the maximum adsorption capacity; both in terms of mg/g. C e is the equilibrium concentration of the adsorbate (mg/L), and K L is the Langmuir adsorption constant (L/mg). Parameter R L is the Langmuir dimensionless factor which indicates the adsorption characteristics. If 0 < R L < 1, it can be concluded that the adsorption was effective. Equation (4) shows how the R L parameter can be calculated (Wei et al. ). In Equation (4), C 0 is the initial concentration in mg/L and K L is the Langmuir adsorption constant (L/mg).
Equation (5) is for the Freundlich isotherm model: In Equation (5), K F is the Freundlich constant, which indicates the adsorption capacity (mg/g). 1/n shows the adsorption intensity and n is the deviation of the data from linear adsorption. If n < 1, the adsorption is chemisorption, and if n > 1, the process is physisorption. If n ¼ 1, the adsorption is linear (Khambhaty et al. ).
Temkin isotherm is given by the following equations: In Equation (6), K T is the equilibrium binding constant (L/g) and β is related to the heat of the adsorption. Equation D-R isotherm is described by the following equations: In Equation (8), q D is the theoretical saturation capacity (mg/g), K is the D-R isotherm constant (mol 2 /J 2 ) related to the mean adsorption energy, and ε is the Polanyi potential (J/mol). Equation (9) shows the relation for ε.
In Equation (9), R is the universal gas constant, T is the absolute temperature, and C e is the equilibrium concentration. To understand whether the adsorption is physical or chemical, the mean energy, E (J/mol), is calculated from Equation (10). If the E value is less than 8 kJ/mol, the adsorption is physisorption and if E is between 8 and 16 kJ/mol, chemisorption had occurred (El-Kamash ).
Studying the adsorption kinetics provides information about the rate of the adsorption and predicts the adsorption-desorption rate of the adsorbate in a solid-liquid system. The kinetic models which are commonly used to describe adsorption are pseudo-first-order, pseudo-secondorder, and intraparticle diffusion. In the pseudo-first-order kinetic model, the main resistance is related to the attachment of the adsorbate to the adsorption sites. Also, it is assumed that the rate of the adsorption sites being occupied is proportionate to the unoccupied sites. The pseudo-firstorder kinetic model is given by the following equation: In Equation (11), q e is the equilibrium adsorption capacity and q t is the adsorption capacity at t; and both are in terms of mg/g. K 1 is the reaction velocity constant (min À1 ).
In the pseudo-second-order kinetic model, like the pseudo-first-order model, the main resistance to the adsorption is present in the attachment of the adsorbate to the adsorption sites. It is assumed that the rate of the adsorption sites being occupied is proportionate to the square of the unoccupied sites. The pseudo-second-order kinetic model can be expressed by the following equation: In Equation (12), q e is the equilibrium adsorption capacity and q is the adsorption capacity at t; both in units of mg/g. Parameter K 2 is the rate constant of the pseudosecond-order model in terms of g/(mg min) (Boparai et al.

).
It is indicated in the intraparticle diffusion model that the adsorption is proportionate to t 0.5 . Equation (13) shows the relations between the parameters in the intraparticle diffusion model: In Equation (13), q t is the adsorption capacity at t in terms of mg/g, K id is the intraparticle diffusion rate constant in mg/(g min 0.5 ), and c is the equation constant which provides information about the boundary layer. The greater the parameter c is, the more the boundary layer affects the adsorption (Boparai et al. ).
In order to better understand different aspects of the adsorption, regarding its spontaneity, being endothermic or exothermic, and how the target ions rearrange toward smaller or greater degrees of freedom, it is important to study the thermodynamic of the process. Thermodynamic parameters such as the standard Gibbs free energy change (ΔG ), the standard enthalpy change (ΔH ), and the standard entropy change (ΔS ) can be calculated according to the changes in the thermodynamic equilibrium constant (K C ), which itself is obtained via plotting ln(q e /C e ) versus q e and the extrapolation of q e to zero. The changes in standard Gibbs free energy is calculated through the following equation, where T is the absolute temperature and R is the universal gas constant.
ΔG shows the spontaneity of the adsorption and the more negative it is; it means the process was more spontaneous.
In order to gain the results of ΔH and ΔS , values of ln(K C ) versus 1/T are drawn in a graph. Then, the ΔH and the ΔS are calculated based on the slope and intercept of the graph and Equation (15) (Vuković et al. ).
RESULTS AND DISCUSSION FTIR, SEM, BET, and zeta potential analyses  group of HDTMA. There was also a negligible peak at 1,473 cm À1 which was assigned to the asymmetric bending state of the quaternary ammonium of the head methyl group of HDTMA (Aroke & El-Nafaty ).
The peaks displayed in HDTMA-Clino occurred in the NaOH-HDTMA-Clino spectrum with slight displacements.
Also, the 3,621 cm À1 bonds were attributed to the stretching vibrations of the hydroxyl group of the loaded surfactant.
The peaks were more intense in clinoptilolite and HDTMA-Clino FTIR spectra than in the one for NaOH-HDTMA-Clino. The decrease in the peak intensities can be due to the decrease in the crystalline characteristic after the alkaline pretreatment (Ates & Akgül ). After the modification with HDTMA, the surface of clinoptilolite seemed to become less heterogeneous. After the alkaline pretreatment with NaOH, the adsorbent showed a slight hierarchical structure. No obvious differences were observed between the SEM images of the modified adsorbents before and after DCF uptake. This can be due to the absence of any chemical reactions after DCF sorption on both adsorbents.
The BET method was used to determine the specific surface area of clinoptilolite before and after modification.
The N 2 adsorption-desorption isotherms at 77 K fitted type IV of the N 2 isotherms which implied the presence of micropores and mesopores in the adsorbent structure.
The BET surface area of clinoptilolite was 14.679 m 2 /g, for HDTMA-Clino, it was 7.329 m 2 /g, and for NaOH- For an 80 mg/L DCF solution, the removal percent was 53.16% after 90 min. The adsorption capacity also followed an ascending trend for 20, 50, and 80 mg/L DCF solutions by increasing the contact time. Also, the adsorption capacity was higher for greater DCF initial concentrations.
The removal percent diagram for NaOH-HDTMA-Clino followed a similar trend as the one for HDTMA-Clino did. adsorption capacity was 4.15 mg/g at 120 min. The adsorption capacity for a 50 mg/L DCF solution was 10.40 mg/g after 120 min. For a DCF solution with 80 mg/L initial concentration, the adsorption capacity reached 11.59 mg/g at 120 min.
These results showed that the equilibrium contact time was 90 min for HDTMA-Clino and 60 min for NaOH-HDTMA-Clino. So, the alkaline pretreatment was effective in this way and it enhanced the adsorption equilibrium time. The other experiments were later done at these contact periods.

Effect of pH
When the pH of the solution is less than the pK a (K a ¼ acid dissociation constant) of DCF, its solubility in the aqueous solution decreases and it starts to precipitate in the solution.
First, the pK a of DCF was investigated and results showed that in pH < 5, DCF starts to precipitate. So, the effect of pH on the adsorption was investigated in the range of 5-10. Figure  adsorption capacity for NaOH-HDTMA-Clino was 8.86 mg/g at pH ¼ 5, and the minimum adsorption capacity was 7.75 mg/g in pH ¼ 10. The maximum adsorption capacity for HDTMA-Clino was 7.77 mg/g at pH ¼ 5, and the minimum adsorption capacity was 6.23 mg/g at pH ¼ 10. When the solution pH is more than the pK a , DCF is present in its anionic form and also its solubility in water increases, so it has a low affinity toward the adsorbent. Effect of the initial concentration  So, the removal percent with both adsorbents decreased slightly and almost negligibly by increasing the temperature.
DCF adsorption capacity with both adsorbents for each initial concentration remained almost the same with temperature changes. This showed that temperature had an adverse, although not considerable, effect on the adsorption and that the process was exothermic. So, 25 C was a more suitable temperature than 35 and 45 C.

Isotherm study
Relevant parameters of the isotherm models are represented in Tables 1 and 2. For DCF adsorption with HDTMA-Clino, the R 2 value of the Temkin isotherm model was 9.972 × 10 À1 at 25 C, 9.854 × 10 À1 at 35 C, and 9.939 × 10 À1 at 45 C.
These R 2 values were greater than the R 2 values of the other isotherm models. Also, the root-mean-squared error (RMSE) value for Temkin isotherm was 9.950 × 10 À1 at 25 C, 2.312 × 10 À1 at 35 C, and 1.467 × 10 À1 at 45 C.
These RMSE values were the least among all the others at each temperature. Thus, for DCF adsorption with HDTMA-Clino, the Temkin isotherm model fitted the experimental data best.
For DCF adsorption with NaOH-HDTMA-Clino, at 25 C, the R 2 value for D-R isotherm was 9.969 × 10 À1 , which was the greatest compared with the R 2 values of other isotherm models at 25 C. The calculated RMSE for D-R isotherm at 25 C was 8.700 × 10 À2 , which was the least compared with the RMSE values of other models at the same temperature. The R 2 value for the Freundlich isotherm model was 9.966 × 10 À1 at 35 C and 9.990 × 10 À1 at 45 C. These  Kinetic study Table 3 shows the kinetic parameters of DCF adsorption with HDTMA-Clino, and Table 4  concentrations, the R 2 values for the pseudo-second-order kinetic model were 9.942 × 10 À1 , 9.913 × 10 À1 , and 9.993 × 10 À1 , which were greater than the R 2 values of the pseudo-first-order kinetic model. Hence, the kinetic of DCF adsorption with both adsorbents was in better agreement with the pseudo-second-order model.   that at greater initial concentrations, the adsorption was mostly affected by the boundary layer and that the intraparticle diffusion was not the sole present mechanism in the process.
Thermodynamic study   To summarize, clinoptilolite is an abundant and inexpensive natural zeolite, which has shown to be effective in DCF adsorption. It has an inherently negative charge and mostly possesses micropores, thus it was modified with a cationic surfactant to have affinity toward DCF anions.
Then, the effect of alkaline pretreatment on the adsorption was also investigated. The equilibrium contact time for DCF adsorption with both HDTMA-Clino and NaOH-HDTMA-Clino was less than 2 h with more than 90% DCF removal, which was a satisfying result. According to BET analysis, although alkaline pretreatment increased the specific surface area and the total pore volume of HDTMA-Clino, it might not be necessary to apply this pretreatment at these concentrations of DCF under these sets of conditions; as the HDTMA-Clino itself adsorbed more than 90% of DCF from the solution and its DCF adsorption capacity improved by a maximum of about 3 mg/g after NaOH pretreatment.

CONCLUSIONS
Clinoptilolite was not effective in DCF adsorption because of its negative surface charge, so it was modified to HDMA-Clino and NaOH-HDTMA-Clino. The zeta potential measurement revealed a positive-surface charge for both modified adsorbents and the BET analysis reported greater total pore volume and mean pore volume for NaOH-HDTMA-Clino. The adsorption equilibrium time with NaOH-HDTMA-Clino was 30 min sooner than HDTMA-Clino. The optimum pH was the least possible pH value before DCF started to dissociate (pH ¼ 5), and the optimum adsorbent amount was 1 g in 100 mL for a 50 mg/L DCF solution. The temperature had a slight adverse effect on DCF adsorption and 25 C concluded to be better than 35 and 45 C. The isotherm study stated a heterogeneous surface for the adsorbents. The E parameter amounts at 25, 35, and 45 C were less than 8 kJ/mol, so the adsorption was physisorption and mainly due to the electrostatic forces. The adsorption was in better agreement with the pseudosecond-order kinetic model, and it was also spontaneous.

DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.